ARGOS: Automated Functional Safety Requirement Synthesis for Embodied AI via Attribute-Guided Combinatorial Reasoning

TL;DR

This paper introduces ARGOS, a framework that uses attribute-guided reasoning with LLMs to generate physically grounded safety requirements for Embodied AI in complex environments, addressing scalability and coherence issues.

Contribution

ARGOS bridges open-ended instructions and physical attributes to generate context-specific safety requirements, advancing automated safety analysis for Embodied AI.

Findings

ARGOS outperforms baselines in identifying long-tail risks.

It produces high-quality, context-specific safety requirements.

The framework effectively grounds LLM reasoning in physical attributes.

Abstract

Ensuring functional safety is essential for the deployment of Embodied AI in complex open-world environments. However, traditional Hazard Analysis and Risk Assessment (HARA) methods struggle to scale in this domain. While HARA relies on enumerating risks for finite and pre-defined function lists, Embodied AI operates on open-ended natural language instructions, creating a challenge of combinatorial interaction risks. Whereas Large Language Models (LLMs) have emerged as a promising solution to this scalability challenge, they often lack physical grounding, yielding semantically superficial and incoherent hazard descriptions. To overcome these limitations, we propose a new framework ARGOS (AttRibute-Guided cOmbinatorial reaSoning), which bridges the gap between open-ended user instructions and concrete physical attributes. By dynamically decomposing entities from instructions into these fine-grained properties, ARGOS grounds LLM reasoning in causal risk factors to generate physically plausible hazard scenarios. It then instantiates abstract safety standards, such as ISO 13482, into context-specific Functional Safety Requirements (FSRs) by integrating these scenarios with robot capabilities. Extensive experiments validate that ARGOS produces high-quality FSRs and outperforms baselines in identifying long-tail risks. Overall, this work paves the way for systematic and grounded functional safety requirement generation, a critical step toward the safe industrial deployment of Embodied AI.